Inspiration

As a team of students, we all share a common problem: finding a place on campus to study or pass the time.

What it does

Spot shows students which tables and spaces on campus are free right now. Raspberry Pi camera units count people inside zones. Staff draw over the camera view, and a live public map updates the moment someone sits down or leaves.

How we built it

Each camera unit runs YOLOv8n person detection on a Raspberry Pi 5 and publicly posts only the counts (no video) to a Next.js app on Vercel. Supabase stores the counts and pushes them over Realtime to the map.

Challenges we ran into

Getting counts to stay steady was harder than detecting people, since a single missed frame would flicker a table's status. We counted each person by their feet, since that's where they actually stand in the camera view, and held any change for 5 seconds before posting it to stop the flicker.

Accomplishments that we're proud of

Person detection is accurate, and users can view spot availability in real time. We also made setting up a new camera unit simple enough that anyone on staff can have one running in a few minutes.

What we learned

We learned how to run real-time computer vision on a small device and keep it in sync with a live web app. We also learned that hardware brings problems no amount of code can fix, like a faulty wire or a broken monitor.

What's next for Spot

We want to put camera units in real campus spaces like the library and food court so students can use it day to day. After that, we'd add trends like typical busy hours so students can plan ahead.